Euresys presents CoaXPress frame grabbers at China Vision show
Highlights at Euresys’ booth will include the latest CoaXPress frame grabbers, the Coaxlink Quad CXP-12, a new four-connection 12-Gbps CoaXPress 2.0 frame grabber. CXP-12 is the top speed of the latest version of CoaXPress 2.0. It operates at exactly 12.5 Gbps! So it is twice the existing speed of the CXP Standard. You’ll also discover the Coaxlink Octo, an 8-connection CXP-6 frame grabber. Both support CustomLogic: custom on-board FPGA processing for Coaxlink series.
Euresys’ Coaxlink Quad 3D-LLE is a four-connection CoaXPress frame grabber with the Laser Line Extraction functionality embedded in the board’s FPGA. It is able to generate 16-bit 3D depth maps directly from the images of a laser line projected onto the object to be inspected.
As the laser-line-extraction process is entirely carried out on board, it does not load the host CPU. The user can choose among several measurement algorithms to accommodate different types of object surface. The measurement precision reaches 1/256 pixel with the Peak and COG (Center of Gravity) algorithms. The maximum measurement speed is 19,000 profiles/s from 1024 x 128 images or 38,000 profiles/s from 1024 x 64 images.
In addition, the 3D depth maps generated by the Coaxlink Quad 3D-LLE are compatible with Euresys’ Open eVision Easy3D image analysis library. Easy3D includes functions such as the calibration and management of point clouds, the generation of calibrated ZMaps and the interactive 3D display of objects with the 3D Viewer. Compared to using off-the-shelf 3D sensors, integrating the Coaxlink Quad 3D-LLE allows the user to build a customized high-performance system, choosing the best camera (resolution, speed), optics, laser and mechanical setup for his application.
EasyDeepLearning is a Convolutional Neural Network-based image classification library. It has been tailored, parametrized and optimized for analyzing images, particularly for machine vision applications. EasyDeepLearning has a simple API and the user can benefit from the power of deep learning with only a few lines of code.
Finally, as specialist of FPGA Cores for machine vision with Sensor to Image products, we will demonstrate the latest versions of the GigE Vision IP Core / MIPI CSI-2 IP Core for FPGA. On the front-end side, we will demonstrate solutions to interface sensors with output according to the MIPI CSI-2 Receiver IP Core interfacing a Sony IMX 274 and MVDK reference design with Xilinx Zynq Ultrascale+.